Professional Documents
Culture Documents
Complex sentences
This chapter is concerned with sentences containing more than one clause, or with major extra-clausal
constituents. Multi-clausal sentences are of primarily two types - those displaying coordination, in which
more than one complete clause occurs at the same level in the syntax; and those displaying subordination,
in which one clause is embedded within another. Subordinate clauses in Kokota have a range of functions,
including modifying nominals (relative clauses) or entire main clauses (adverbial clauses), or acting as a
main clause argument (as subject, complement or adjunct). In addition, the chapter examines a number of
minor constituent types which occur outside the main clause but are not themselves coordinated main
clauses, includes recapping constituents (demonstratives and reduced clauses), and 'be thus' clauses based
on the verb -u 'be thus'.
11.1 Coordination
The particle ge occurs either clause initially or clause finally, coding a sequential relationship between the
marked clause and another constituent. When the particle occurs clause finally, it marks the event coded by
the clause as preceding the event coded by the next clause:
More commonly the particle occurs clause initially. In this position it marks the event coded by the clause
as following the event coded by the preceding clause:
276
ge g#-e ag#e lehe-n#a n-e-ke-u
SEQ NT-3.SBJ go die-IMM RL-3.SBJ-PRF-be.thus
and then he died, that's how it was.
The relationship between ge and the clause it marks is thus iconic: it follows clauses marked as preceding
and precedes clauses marked as following. The order of the two clauses is also typically iconic, as the
examples in (11.1) and (11.2) illustrate. However, while the iconic relationship between the particle and the
clause it marks is obligatorily reflected in their structural relationship, the ordering of the two clauses only
tends to be iconic - the reverse order is also possible. In (11.3) the clause which represents the event which
occurred second in time actually precedes the clause representing the event which occurred first in time.
However, the clause representing the second event is still marked with ge in its iconic clause initial
position:
The fact that the sequencing particle ge occurs sentence initially in (11.3), not between the coordinated
clauses, demonstrates two facts. Firstly it demonstrates that the two clauses are indeed coordinated in a
single sentence, since the semantic relationship between the clauses cannot result from their order alone.
Secondly it demonstrates that ge is not a conjunction at this clausal level: since the particle does not occur
between the clauses it cannot be conjoining them. The particle marks individual clauses in a way that
conveys information about the temporal relationship between the event coded by the marked clause and
other events. No conjunction exists in these coordinated structures.
In all the examples in (11.1) to (11.3) both coordinated clauses are positive. However one of the sequenced
events can be negative:
Constituents related at the clausal level are not limited to pairs of clauses. Other sentence level constituents
may be related sequentially to a clause. Temporal constituents may be related to a clause in this way. In
(11.5)a. a temporal locative occurs with a ge-marked clause, indicating that the time coded by the temporal
locative will precede the event coded by the clause. In (11.5)b. a temporal interrogative occurs. In (11.5)c. a
local noun phrase occurs with legu 'behind' with its temporal meaning. This sentence level adverbial phrase
is related sequentially to the clause by ge.
b. niha-na ge da #
lao-na buala
when-thatN SEQ 1INC.SBJ go-IMM PNLOC
When will we go to Buala?
277
c. legu-na toka kave ana gita g#azu ana
behind-3SGP chop descend thatN weINC wood thatN
After we have cut down that tree
As discussed in 11.3, Kokota discourse style employs frequent recapping, often with a demonstrative
referring to the event coded by the previous clause, or a prepositional phrase consisting of a clausal
demonstrative (ka-t-au-), also referring to the event coded by the previous clause. With both of these
recapping strategies ge commonly occurs marking the main clause:
Ge also occurs introducing the second clause in periphrastic manner and cause interrogatives. This is
discussed in 10.2.3.
Ge has a variant form age, which occurs in clause initial position only. Age primarily occurs when the
preceding clause itself has a clause final ge. This dual sequencer marking occurs frequently. The first of the
sequentially related clauses has ge in final position, marking the event coded by the clause as preceding a
subsequent event. The second clause has age in initial position, marking the event coded by the clause as
following a preceding event. When dual sequencer marking occurs, the clause initial second clause
sequencer must have the form age.
278
b. n-e la de-deke-u sini ge
RL-3.SBJ go RD-step-PRG FOC SEQ
He stepped and then
This dual sequencer marking occasionally occurs with other sentence initial constituents such as recapping
PPs:
(11.8) ka-t-au-ao ge
LOC-SB-exist-thisT SEQ
From that
Occasionally age occurs clause initially in place of ge in clauses where the preceding clause does not have a
final ge:
In addition to its clause marking sequencer function, the particle ge occurs below the level of the clause as a
conjunction. With this function there is no sequential sense, the particle simply conjoining phrasal and word
level constituents. Only constituents of the same constituent class may be conjoined in this way. In
(11.10)a. two adnominal post-head core modifiers (in this case personal names) are conjoined within a
single NP core and modify a single nominal head (mane 'man'). In (11.10)b. two verbs are conjoined in a
single predication.
Phrasal constituents are also conjoined by ge. In (11.11) a number of different NP types are conjoined. In
(11.12) two locative obliques are conjoined.
The conjunction n#a 'but, instead' conjoins clauses only. It occurs clause initially in the second of the two
conjoined clauses:
The n#a marked clause may occur as the first clause in an utterance, but only if the situation for which the
clause is an alternative has just been established in the discourse, as with speaker B's response in (11.14):
Although n#a may be sensibly glossed as 'but', it does not correspond exactly to that English conjunction.
The form in fact introduces a clause coding a situation which is presented as a contrast to the situation
coded by the preceding clause. This contrastive function often gives the conjunction a sense more akin to
English instead. In the piece of text in (11.15) the conjunction opening clause 2 relates that clause to the
preceding clause with the sense that despite the situation coded in the preceding clause, the situation coded
in clause 2 pertains. Specifically, despite the fact that the piece of taro was brought by the participants
mentioned, and contrary to the positive expectations implicit in that, the speaker is dying from that piece of
taro. The conjunction introducing clause 3 then contrasts the situation coded in that clause with that coded
in clause 2.
280
n#a ne lehe-ni a-hi ara
but RL die-3SGO thisT-EMPH I
but I'm dying from it,
The situation coded in the second (n#a marked) clause may be contrasted with a situation which is not
overtly expressed in the preceding clause, but is implicit in it. In (11.16) the addressees live at Putuo. The
situation coded in the second clause is presented in contrast to the addressee's living in Putuo, not to the
opinion that Putuo is not much good:
Events which relate closely to each other are often expressed by clauses which are not coordinated, but are
merely adjacent to each other in the discourse and represent separate sentences:
However, smaller constituents which occur within a single clause may be coordinated without an overt
coordinator, especially constituents marked with a constituent marker such as bo 'contrastive' or ba
'alternative' (which are not themselves conjunctions, as discussed in 9.8). In (11.18)a. two verb complexes
are coordinated, in (11.18)b. two NPs:
The form n-e-u consists of the verb -u 'be thus', preceded by the realis and 3SG subject markers n-e. This
occurs very commonly, with a range of functions, the most common of which is as a tag clause (see 11.4).
A further function is to introduce a constituent and conjoin it to a preceding constituent. In this sense the
form appears to be undergoing a process of grammaticalisation in which it is becoming a conjunction.
The form occurs commonly introducing a clause, where the situation coded by the second clause is
presented as cooccurring with the situation coded by the first clause.
281
(11.19) kai bo au ia pirisi hugo hebala
later CNT exist theSG priest PN PN
Later there was the priest Hugo Hebala.
It is not entirely clear that the clauses in lines 2 and 3 of (11.19) are even conjoined and form a single
sentence, in the sense that as a single complex sentence the two clauses would have some psychologically
real relationship closer than that of clauses which are merely adjacent in the discourse. By introducing the
clause in line 3, n-e-u appears to mark that clause as coding a situation that is associated in some
semantically close way with the situation coded by the preceding clause, in this instance that the events
occurred concurrently. In that example it is n-e-u alone which indicates the concurrence captured by the
bracketed element of the free translation. The literal translation of n-e-u is something along the lines of 'it
is/was thus' or 'be thus'. This meaning is readily apparent in the many instance where a clause is introduced
by n-e-u but clearly does not form a single sentence with a preceding clause, as speaker B's response in
(11.20) illustrates:
B. n-e-u hae
RL-3.SBJ-be.thus where
Where? [lit. Be thus where?]
The association with a situation coded by a preceding clause does not necessarily involve concurrence. The
relationship is frequently sequential:
The semantics of the clauses in (11.21) more readily give a conjunction-like appearance to neu. However,
this is more so when the form occurs between constituents smaller than a clause. In (11.22) the first
instance of neu apparently conjoins individual verbs or perhaps verb complexes. However, the presence of
the limiter-marked second occurrence of neu undermines a conjunction analysis here, suggesting as a more
literal translation of the clause something like "dancing, likewise playing, they were just like that there".
However, in (11.23) neu occurs between two verb complexes, apparently both marked by the same sentence
final tag clause n-a-u 'I am/was thus'. Here the behaviour of the form is more strongly conjunction-like:
282
(11.23) ne kapo no-g#u n-e-u marh-i-di pau-g#u-de
RL feel.cold GP-1SGP RL-3.SBJ-be.thus feel.pain-TR-3PLO head-1SGP-theseR
I'm cold [ie. feverish] and my head hurts,
n-a-u
RL-1.SBJ-be.thus
I'm like that.1
The most conjunction-like appearance of neu, however, is when it occurs between NPs. In (11.24)a. the
form conjoins two subject NPs in a subordinate clause, in (11.24)b. it conjoins two NPs governed by a
single preposition.
The presentation of alternatives does not involve a conjunction comparable to the English or. Instead, this
function is performed by the marking of each alternative constituent with the constituent modifier ba
'alternative'. This may mark constituents of any size from individual words up to complete clauses. This
particle is not a conjunction, and is discussed in detail in 9.8.1.
11.2 Subordination
Several clause types occur subordinated within a main clause. Some subordinate clauses occur immediately
governed by the sentence node with the function of modifying the entire main clause, while others function
adnominally. Others function as arguments of the main clause predication, or as nominal predicates in non-
verbal clauses.
In general, subordinate clauses with any of these functions fall into two categories - realis and irrealis.
Realis subordinate clauses have no subordinating particle. Irrealis subordinate clauses are governed by the
subordinating particle ta. This latter category includes most conditional clauses.
Several minor types of adverbial subordinate clauses also exist, each with their own formal characteristics.
Most subordinate clauses fall into two types - those which code a realis event and those which code an
irrealis event. The former do not involve any subordinating particle, while the latter do.
The modal characterisation of events in subordinate clauses in part reflects the way those events would be
treated if they were expressed by a main clause. The basis of the main clause classification of events as
realis or irrealis is discussed in 8.5.2.3. However, the range of subordinate clause predications which are
1
Note that pau ‘head’ may be plural even when referring to a single individual’s head.
283
treated as irrealis is considerably wider than those treated as irrealis in main clauses. Any real event which
actually occurred before the moment of speaking, or is actually occurring at the moment of speaking, is
treated as realis. In subordinate clauses all other events are treated as irrealis. The prototypical non-real
event is one which is located at a time after the moment of speaking, in other words one which has yet to
occur. This prototypical distinction is neatly reflected in relative clauses in terminology for divisions of
time:
b. ka wiki ta mai-ne
LOC week SB come-thisR
next week (lit. this week that will come)
In (11.25)a. the week referred to is in the past, and its going has actually occurred. As such the relative
clause coding the event is realis, with a realis auxiliary and no subordinating particle. In (11.25)b. the
coming of the week has yet to occur. As such the event is not yet real and it is treated as irrealis and the
relative clause has no auxiliary. Instead the subordinating particle ta is present. In effect, an auxiliary and
the subordinator are in complementary distribution in subordinate clauses. The absence of an auxiliary in
irrealis subordinate clauses correlates with the omissibility of the person auxiliary in irrealis main clauses,
and the (crosslinguistically unusual) status of irrealis as the unmarked of the two modal categories.
Any subordinated positive active predication is realis. While the examples in (11.25) are useful from a
contrastive point of view, (11.25)a. does not reflect a prototypical use of a realis subordinate. The following
(bracketed) relative clause is more typical:
Irrealis subordinate clauses cover a much wider range of events. These include events which, while positive
and active, have yet to occur. Such events are treated as irrealis in subordinate clauses, as they are in main
clauses:
However future events are not the only events which are expressed as irrealis in both main and subordinate
clauses. Habitual events, while being positive events, are coded as irrealis as they are not actual specific
events, as the main clause in line 3 of (11.28)a. illustrates. The act of referring to an entity by the word
which a language assigns to that entity is inherently habitual so is also treated as irrealis, as in (11.28)b.-c.:
284
[ta la hod-i-di-re gai]
SB go take-TR-3PLO-thoseN weEXC
that we take.
In addition to habitual events, hypothetical events are treated as irrealis, with subordinated clauses
expressing hypothetical events coded irrealis:
Subordinate clause events located in the past or present are realis if positive. However, the non-occurrence
of an event is treated as irrealis regardless of the temporal frame of the event. Thus past counterfactual
((11.30)a.) and present counterfactual ((11.30)b.) subordinated clauses are irrealis.
Note that the subordinate clauses in (11.30) are not negative. Instead they express positive events.
However, the wider context of the main clause indicates that these events did not occur, and thus, despite
expressing positive events located in the past or present, the clauses are coded as irrealis.
Like habitual events, states have a validity that holds beyond individual temporal locations or modal status.
However, in main clauses states may be coded as realis if the state actually exists or existed at a particular
point in the past or present. Subordinate clauses expressing states vary in their modal coding. In relative
clauses states are always coded as irrealis. This is as true of temporary states, as in (11.31)b., as it is of
permanent states ((11.31)a.):
285
(11.31) a. la-i bla kaike tu-turi [ta puku] bl-ago
go-3SGO LMT one RD-tell SB be.short LMT-youSG
Just tell a story that's short.
However, other subordinate clause types resemble main clauses in that they code real states as realis. For
example the complement clause in (11.32) is realis:
Like states, relationships such as knowing or possessing have a non-specific quality in reality or temporal
terms. However, like states, while main clauses expressing such relationships are coded realis, subordinate
clauses must be irrealis:
Like states, existential clauses are realis as main clauses but irrealis as subordinate clauses:
The use of the subordinator ta with the existential verb au has given rise to formulaic clausal
demonstratives such as t-au-na 'that (nearby)' (lit. 'that which is that') and t-au-de 'these (within reach)' (lit.
'that which are these'). These clausal demonstratives are discussed in 4.1.3.3.
Irrealis subordinate clauses are typically marked with the subordinator ta, as discussed above. However,
where the context makes clear the irrealis status of the subordinate clause, the subordinator is occasionally
omitted. This occurs very commonly when the subordinate clause is marked with the purposive marker
mala. The prototypically unrealised nature of intended events gives purposive subordinates a prototypical
irrealis status. As discussed in 11.2.7, this licences the omission of the subordinator. However, the
subordinator may occur in such clauses, as (11.35)a. illustrates. Less commonly, omission of the
subordinator occurs with other irrealis subordinate clauses, typically where the event coded by the
subordinate clause is located in the future within the temporal frame of the main clause. When the
286
subordinator is omitted an irrealis auxiliary may occur. As discussed in 8.5.2.2, irrealis is realised by zero
marking, contrasting with marked realis and neutral categories. The auxiliary thus consists only of the
person indexing vowel. This auxiliary may occur when the subordinator is omitted, as in (11.35)b. In main
clauses irrealis auxiliaries tend to be omitted when there is no ambiguity about the identity of the subject or
actor. This is also true of subordinate clauses. Consequently an irrealis subordinate clause from which the
subordinator has been omitted may have no auxiliary, as in (11.35)c.
As discussed in 9.2, the pragmatically unmarked clause constituent order in the language is VAO or VS.
However, main clauses allow the preverbal topicalisation of any argument. In addition, a clause final focus
position exists. Subordinate clauses of all types have the same pragmatically unmarked constituent structure
as main clauses. However the pragmatically marked possibilities differ from those of main clauses.
The topicalisation possibilities for relative and complement clauses differ between zero-marked clauses
and those with the subordinator ta. However, all zero-marked clauses allow the same possibilities
regardless of whether they are functioning as relative or complement clauses, as do all ta marked clauses.
Clauses marked with the subordinator ta do not allow any argument to occur in preverbal topicalised
position. This applies to all ta marked clauses regardless of main clause function.
In zero-marked relative and complement clauses a preverbal topicalised argument is possible, but is
ergative - only actors (ie. the subjects of transitive predications) may be topicalised, as (11.36)b.-c. illustrate
for relative and complement clauses respectively. Intransitive subjects are precluded from occurring
preverbally, even if unergative ((11.36)d.), as are objects ((11.36)e.-f.). The pragmatically unmarked
structure is shown in (11.36)a.
Note that although manahagi 'want' is potentially ditransitive, the ago 'youSG' in (11.36)c. is within a direct
object complement clause, indexed by the third singular agreement enclitic on the main clause verb, not a
direct object separate from an indirect object complement clause, as would be the case if the main clause
agreement enclitic was second singular:
As the controlled argument may not be overtly realised in relative clauses (as discussed in 11.2.4),
topicalisation is only possible when the subordinate actor is not the controlled argument.
Like main clauses, both ta marked and zero-marked relative and complement clauses allow an argument in
clause final focus position. However in main clauses a focussed argument is marked with the focus particle
si. In subordinate clauses of all types this focus particle may not occur. Instead, in relative and complement
clauses a focussed argument occurs in clause final position without si. This occurs infrequently. More than
one argument must be present in the subordinate clause. Moreover, intransitive subjects and objects occur
in clause final position unless an oblique is also present. Consequently it is typically a transitive actor which
is focussed in subordinate clauses, as (11.38) illustrates for zero-marked and ta marked relative and
complement clauses:
However, relative and complement clause focussing is not limited to actors. In (11.39), for example, an
intransitive subject occurs in clause final position, following an oblique:
288
(11.39) ara manahagi-ni [ta mai ade ago]
I want-3SGO SB come here youSG
I want that you come here.
Realis subordinate clauses occur as adnominal modifiers identifying or characterising the head nominal on
the basis of an event in which the participant coded by the head nominal took part, or a state which applies
to that participant.
The behaviour of relative clauses within NP structure is discussed in 4.3.2.2.3.6. As discussed in that
section, two types of relative clauses exist: reduced and full. Reduced relative clauses consist only of the
subordinator ta plus a single stative verb, and occur within the NP core. Nothing further needs to be added
here regarding reduced relative clauses other than that they may modify any nominal main clause argument.
Full relative clauses are NP outer modifiers. The rest of this section deals with full relative clauses.
Any main clause argument may be modified by a relative clause. In (11.40)a. the main clause actor (ie.
transitive subject) is modified, in (11.40)b. an unergative intransitive subject, in (11.40)c. an unaccusative
subject, in (11.40)d. an undergoer, in (11.40)e. an oblique, and in (11.40)f. a possessor:
289
11.2.3.2 Relative clause argument roles
The participant expressed by the main clause argument which is modified by the relative clause (ie. the
coreferential argument) may have any grammatical relation in the relative clause. It may have the same
relation in both clauses, as in (11.40)a., where the coreferential argument is an agent and transitive actor in
both clauses. In (11.40)b. the coreferential argument is a [+A] argument in both clauses, but has slightly
different grammatical relations: in one it is a transitive actor and in the other an unergative subject.
Similarly in (11.40)c. the coreferential participant is a [-A] argument in both clauses, an unaccusative
subject in the main clause and a theme in the relative clause. Alternatively, the argument may have
completely different roles and relations, as in (11.40)d. and e., where the coreferential argument is an
oblique in one clause and an undergoer in the other. The fact that both these roles are not [+A] is not
significant - a participant may be the [+A] argument of one clause and the [-A] argument of the other. In
(11.40)f. the main clause oblique is [+A] in the relative clause (an unergative subject), while in (11.41) the
[-A] main clause undergoer object is [+A] relative clause actor:
The freedom of participants to function in any role in both main and relative clauses creates the potential
for ambiguity, as in (11.41). Here neither the actor nor undergoer of the transitive relative clause are overtly
realised. Since the two main clause participants are also the two relative clause participants, and both are
third singular, some means of distinguishing between the two participants in the relative clause is
necessary. However, this means need not be linguistic.
Where no overt arguments are present in the relative clause the ambiguity is not resolved syntactically: the
dog cannot be assumed to have the same role in the relative clause as it does in the main clause, as it could
well have a different role (as it does in (11.41). In situations like this ambiguity is resolved pragmatically
and semantically. In (11.41) the dog would normally be assumed to be the actor of the relative clause
because dogs typically bite, and chiefs typically do not. If a semantically anomalous event was being
described, the unusual role assignment would require an overt realisation of the arguments, in which case
constituent order would resolve the ambiguity. Equally, where either participant could readily perform
either role, overt mentions allow constituent order to resolve the ambiguity. In (11.42), for example, no
ambiguity is possible as the unmarked VAO constituent order dictates the reading.
290
11.2.3.3 Relative clause argument role tendencies
A relative clause may modify a main clause argument of any argument role. However, unelicited data
reflects various tendencies. Relative clauses on main clause undergoers and intransitive subjects occur
commonly; and on obliques slightly less so, while relative clauses on actors occur very infrequently.
The role in the relative clause of the controlled argument shows similar tendencies: by far the most
common controlled arguments are relative clause intransitive subjects. Undergoers are less common,
obliques still less so, and coreferential actors are rare.3
The constituent order possibilities for subordinate clauses are discussed in 11.2.2. However, a constraint
applies to relative clauses that does not apply to other subordinate clause types: the relative clause argument
which is coreferential with its main clause head is subject to control by the head and may not be overtly
realised except by agreement in the subordinate verb complex. An NP realising the controlled argument
may not occur.
As the coreferential argument is controlled, a relative clause cannot have all its arguments specified other
than by agreement. In addition, relative clauses are subject to the same tendency as other clause types to
realise highly activated participants by zero anaphora (see 9.3). Consequently relative clauses frequently
consist of the verb complex only. Where an argument is realised, it is typically either a transitive actor or an
undergoer in a clause where the head fulfils the other core transitive role, or is an oblique. In (11.43)a. the
controlled argument is the object and in (11.43)b. the actor. In (11.43)c. it is an oblique.
Since intransitive clauses have only one core argument, intransitive subjects are very rarely overtly realised
in relative clauses. This is only possible where the head functions as a relative clause oblique. Equally,
transitive relative clauses with both core arguments overtly realised occur very infrequently, and only in the
same circumstances:
n-e dou
RL-3.SBJ be.big
was big.
Relative clauses with more than one overt argument are also possible where the arguments are both
obliques:
[n-e-ke au-re
RL-3.SBJ-PRF exist-thoseN
who lived
Like controlled core arguments, controlled relative clause obliques are not overtly realised. When this
occurs the entire prepositional phrase has a zero realisation, even if in the main clause the coreferential
argument is not also an oblique. In (11.44) a main clause unaccusative subject is modified by a relative
clause in which the coreferential participant functions as an instrument. Instruments are normally realised
by prepositional obliques with the preposition ka as head (as discussed in 7.7.1.6). However, when a main
clause core argument occurs as a controlled relative clause oblique, as in (11.44), no preposition occurs in
either clause. This is also illustrated in (11.46), where the main clause object controls a relative clause
instrument which would otherwise be realised within a ka prepositional oblique:
Relative clauses are potentially recursive, with arguments of one relative clause themselves eligible to be
modified by a relative clause. In (11.47) the object arguments of the relative clause in line 2 are themselves
modified by the relative clause in line 3:
Full relative clauses optionally contain a cliticised demonstrative agreeing with the controlled argument.
This demonstrative attaches to the verb complex. Where a relative clause is intransitive, the verb complex
may be marked with an enclitic agreeing with the subject of that clause (unergative or unaccusative):
292
[ta kave-na] ka maneri k-au toa-na
SB descend-thatN LOC they LOC-exist fort-thatN
who came out, of those in the fort.
In transitive relative clauses the demonstrative enclitic attaches to the postverbal agreement marker. Where
the controlled argument is the relative clause actor the demonstrative agrees with that argument. This can
be seen in (11.49)a., where the relative clause object is plural and the actor singular. Where the controlled
argument is the object, as in (11.49)b., the postverbal agreement marker and the demonstrative enclitic form
a sequence which agrees with both the person and number of the undergoer, and its demonstrative category.
[n-e-ke la hod-i-ri-ro]
RL-3.SBJ-PRF go take-TR-3PLO-thoseNV
she had picked.
The demonstrative may also agree with a controlled oblique. In (11.46) the demonstrative enclitic -ro 'those
(not visible)' agrees with the controlled instruments.
No evident formal or syntactic bases motivate the presence or absence of a demonstrative enclitic. Instead
the motivation appears to be pragmatic - the demonstrative is used to facilitate referent identification.
Speakers have the option of employing this strategy if they judge it useful on a clause by clause basis.
However, while elicited relative clauses often do not contain a demonstrative enclitic, that is not an accurate
reflection of language use. In almost all unelicited relative clauses in the corpus a demonstrative enclitic is
present.
Subordinate clauses may function as arguments of a main clause. As events or states cannot be volitional
entities, arguments realised by subordinate clauses are limited to non-volitional semantic roles. One effect
of this is that argument clauses may not function as the agent of a transitive clause, or as an unergative
subject. Beyond that, they may occur with any grammatical relation.
While an argument clause may not function as agent, it may occur as the actor of a transitive predication,
with the semantic role of force. In (11.50) the clause frin#he heta 'work hard' is the actor of the transitive
causative predication, and is marked with the demonstrative ine 'this (reachable).
293
While subject clauses are typically marked with a demonstrative, this is optional:
Force argument clause actors have a superficial resemblance to nominalised adverbial contextual
subordinate clauses (discussed in 11.2.5.1). In (11.52)a. the initial argument clause is simply the context in
which the speaker's thirst occurred. In (11.52)b., however, a virtually identical clause is the force actor of
the main clause predication. While the argument clause is virtually identical in both sentences, the main
clause in (11.52)a. is intransitive, with the speaker as subject, as the preverbal agreement indicates. In
(11.52)b. the main clause is transitive, with a causative marked predication and the speaker as object
(indexed postverbally).
Clauses which function as the subject of an unaccusative intransitive predication typically express an event
which the main clause comments on in terms of its state or some characteristic associated with it:
The use of a main clause with a subordinate clause as subject occurs frequently in exposition as a recapping
device indicating the progression of events, often indicating the completion of one stage in a sequence of
events:
As with force subordinate clause actors, subordinate clause unaccusative subjects may or may not be
marked with a demonstrative, as (11.53) and (11.54) illustrate. In most elicited sentences with a subordinate
clause unaccusative subject a demonstrative is present, while in most unelicited sentences the clause is not
marked by a demonstrative.
A number of verbs subcategorise for a sentential complement as either direct or indirect object. These
include:
294
(11.55) a. tahe 'tell'
b. snakre 'allow'
c. lubati 'allow'
d. fa noto 'cause to stop'
e. fa teo 'cause to be not'
f. fa nag#r-i 'cause to be blocked'.
g. gato 'think about'
h. manahagi 'want'
i. g#onu 'be insensible/not know'
j. gato-g#onu 'forget'
k. fa nhigo 'cause to be finished'
l. ooe 'say'
Of these, the verbs in (11.55)a.-i. subcategorise for either one or two complements. Where only one
complement occurs it may be an NP or a complement clause. Where two complements exist, the direct
object is always an NP, and the indirect object a complement clause. The verbs in (11.55)j.-l. subcategorise
only for a single complement, which may be an NP or a complement clause. The semantic possibilities for
direct and indirect complement clauses with these verbs fall into four groups:
Table 11.1: The semantics of direct and indirect object complement clauses.
With the verbs in group A the main clause actor is not coreferential with the subject/actor of the
complement clause. The subject/actor of the complement clause may be overtly realised within the
complement clause, as in (11.56)a. Alternatively, the subject/actor of the complement clause may be
realised as the direct object of the main clause, as in (11.56)b., in which case the complement clause is the
indirect object. In this second alternative the complement clause subject/actor is coreferential with the main
clause direct object and may not be overtly realised in the complement clause.
Group A verbs with indirect object complement clauses require coreference between the main clause
undergoer and the complement clause subject/actor. The effect of this is that indirect object complement
clauses do not allow an overt subject/actor within the complement clause itself.
295
With the verbs in group C the complement clause direct object realises an event which the main clause
actor performs. The complement clause subject/actor must therefore be coreferential with the main clause
actor. As with coreferential main clause nominal direct objects in group A verbs, the coreferential argument
may not be overtly realised. Consequently, with group C verbs the complement clause subject/actor may
never be overtly realised:
The desiderative verb, manahagi, behaves in the same way as both group A and group C verbs. As with
group A verbs, the actor of manahagi may want another participant to act. This may be expressed either
with a direct object complement clause containing the complement clause subject/actor, as in (11.58)a., or
with an NP direct object realising the other participant, and an indirect object complement clause realising
the event, as in (11.58)b. Again, in this situation the indirect object complement clause subject/actor is
coreferential with the main clause direct object and may not be overtly realised.
However, manahagi also behaves in the same way as the group C verbs, with a direct object complement
clause realising an event which has as its actor the same participant as the actor of the main clause. Again
the complement clause subject/actor is coreferential with the main clause actor and cannot be overtly
realised:
One further verb, ooe 'say', behaves like the verbs in group A, except in two important respects. Like group
A verbs it may occur with a direct object NP and an indirect object complement clause, with the
subject/actor of the complement clause coreferential with the main clause direct object and not overtly
realised:
(11.60) manei n-e ooe-nau ara [ta mala tazi-ni-na sote ana]
he RL-3.SBJ say-1SGO I SB PURP keep-3SGO-thatN shirt thatN
He said to me to keep that shirt.
However, unlike group A verbs, it appears that ooe may not occur with a complement clause direct object.
Moreover, with ooe there is no restriction that the complement clause subject/actor be coreferential with the
4
Note that in this example the verb g#onu occurs in a reduced form comprising a single word with
postverbal aspectual possessor marking. This reduction, discussed in 8.6.1, also commonly occurs with
mhano 'be afraid'.
296
main clause direct object. As these two arguments may not be coreferential, it is possible for the
complement clause subject/actor to be overtly realised:
Complement clauses occur in the pragmatically unmarked main clause position for the argument they
represent. In bivalent main clauses the complement clause occurs in postverbal position, following the actor
if that argument is also realised postverbally, as (11.59) illustrates. In trivalent clauses, the complement
clause occurs postverbally, following both the actor, if present, and direct object, as (11.60), (11.61) and
other examples above illustrate.
Subordinate clauses may function as an oblique adjunct, governed by the preposition ka. In this context the
clause is obligatorily marked with a demonstrative or article. Being nominalised in this way, possessor
indexing enclitics may also occur with these clauses. Oblique subordinate clauses typically mark a location
of some kind. This may be an event because it is a location in a series of events:
The oblique may also represent a physical location associated with an event, either characteristic of the
location, as in (11.63)a., or temporarily associated with it ((11.63)b.):
297
[ka-[n-e-ke hure-ri-ro ira tilo tomoko [n-e zaho kutare]]]
LOC-RL-3.SBJ-PRF carry-3PLO-thoseNV thePL three war.canoe RL-3.SBJ go mud.shell
to where they had carried the three canoes, which had gone like mudshells.
In addition to functioning as main clause oblique adjuncts, nominalised clauses occur with a number of
other minor functions. This includes as possessor. In (11.64) the purpose of the coming of Gobilologu is the
possessor of the road in the main clause:
A nominalised clause may also modify a pronoun - not as a relative clause, but in the same way that an NP
may modify a pronoun as discussed in 4.4.1.4. In (11.65) the pronoun gai 'we (exclusive)' (itself the object
of a nominalised clausal oblique) is modified by a nominalised phrase marked with the article ira 'the
(plural)':
As a result of the range of functions open to subordinate clauses, a subordinate clause may occur embedded
within another subordinate clause. Relative clause recursion has been discussed in 11.2.3.5. Embedding
also occurs with argument and adverbial subordinate clauses. The range of possibilities is wide. In (11.66),
for example, a nominalised clause oblique itself has a complement clause object:
A number of adverbial functions may be performed by subordinate clauses. With most the clause is
governed by a constituent which is an adjunct of the main clause head. These constituents include the
contextualising nouns gu- and nafu-, the temporal local nouns legu- 'behind' and g#ilu 'inside', and the
temporal locative gilai 'until'. Some contextual clauses, along with one further adverbial clause type, the
affective clause, are not governed by another constituent, but are governed directly by the sentence head.
5
The verb fa-mane-mane has the morphological structure indicated by the morpheme gloss, but the
lexicalised meaning of 'be very happy'.
298
11.2.5.1 Contextual clauses
A nominalised subordinate clause may indicate the context in which the main predication holds. In (11.67)
the initial clause provides the context in which the speaker's thirst occurred:
Contextual clauses of this kind are nominalised, typically by a cliticised demonstrative, and are neither
realis nor irrealis, but are related temporally only to the main clause event. In (11.67) the event expressed
by the contextual clause has occurred, so would be coded realis as a main clause, while in (11.68) the event
has yet to occur, so would be coded irrealis, however the two are structurally identical.
Two contextualising nouns, gu- and nafu, are discussed in 5.6. As contextual nouns they function in an
identical manner. However, nafu also has the root meaning of 'base'. As it is in the process of being
grammaticalised as a contextual noun, and speakers still identify the root meaning when it occurs with its
contextualising function, it is glossed as 'base'. Gu-, on the other hand, occurs only as a contextualiser, and
is glossed as 'CNTX'. Gu- and nafu are possessor indexed to a constituent which provides contextual
information about the event coded by the main clause. The indexed constituent (the complement of the
possessor indexing) may be an NP. However, it may also be a subordinate clause expressing an event which
provides the context for the main clause event:
(11.69) a. ara n-a babao no-g#u [gu-na [n-a frin#he heta fakamo]]
I RL-1.SBJ be.tired GP-1SGP CNTX-3SGP RL-1.SBJ work be.strong always
I'm tired because I always work hard.
Clauses subordinated by gu- and nafu have identical formal characteristics as main clauses. They are not
subject to the constituent structure restrictions discussed in 11.2.2, and the subordinator ta does not occur.
Modal auxiliaries occur as with main clauses (as (11.69) illustrates), and any or all arguments may be
overtly realised, even when they are coreferential with main clause arguments. All main clause constituent
order possibilities occur in contextual clauses. As (11.70) illustrates, arguments with any grammatical
relation can occur in preverbal topic position, including actors, unergative and unaccusative subjects,
objects, and obliques.
299
b. ara n-a fahega
I RL-1.SBJ be.happy
I am happy
Clauses governed by gu- and nafu may also locate an argument in clause final focus position. The focussed
argument may be marked with the focal particle si, as in main clauses, though the focal particle may also be
omitted (as it must be in relative and complement clauses):
Clauses subordinated by gu- and nafu may have a non-verbal predicate. In (11.72) an equative clause is
governed by nafu:
300
(11.72) n-a no-mai ag#e mhemhe-ni-u gai
RL-1.SBJ GP-1.SBJEXCP go be.difficult-3SGO-PRG weEXC
We are finding it hard
Conjoined clauses may be subordinated to gu- or nafu, in which case the possessor indexing on the
contextual noun is plural, reflecting the plural contextual bases expressed by the conjoined clauses:
Clauses governed by gu- and nafu typically occur finally within the main clause. However, they may occur
main clause initially, as (11.70)e. illustrates.
Two local nouns have temporal as well as spatial locative functions. These are legu 'behind, after' and g#ilu
'inside, within, during', and are discussed in 5.4.1. Both typically have a nominal complement. However,
both may also have a subordinate clause as complement. In this situation the local noun carries third person
possessor indexing agreeing with the subordinated clause. When the complement is a subordinate clause
only a temporal reading is possible.
Clauses which are subordinated by legu are nominalised by means of a demonstrative or possessor indexing
marking the subordinate verb complex. These clauses express an event which precedes the event coded by
the main clause.
Clauses governed by legu typically occur main clause initially, iconically reflecting the actual temporal
sequence of the events. However, this is a tendency, not a categorical restriction, as main clause final legu
subordinates occasionally occur:
301
(11.75) ...huhuran#i au-re keha-re n-e-ke-u gai,
PNLOC exist-thoseN NSP-thoseN RL-3.SBJ-PRF-be.thus weEXC
...some of us lived at Huhurangi
Local nouns typically occur immediately governed by the sentence head. However, they may instead be
governed by an intervening preposition. This applies when the local noun complement is a subordinate
clause as much as when it is an NP:
Nominalised clauses governed by legu allow only the pragmatically unmarked constituent order VS/VAO.
No pragmatically marked constructions such as preverbal topicalisation or clause final focussed arguments
are possible.
The behaviour of subordinate clauses governed by g#ilu 'inside' is not fully understood. They do not appear
to be nominalised:
The particle gilai 'until' introduces a subordinate clause which expresses an event marking the end of the
event expressed by the main clause. In positive main clauses gilai indicates that the main clause event
finishes at a point in time coinciding with the occurrence of the subordinate clause event:
6
The verb boro refers to a period after the birth of a child when the mother and the infant remain together
indoors in close physical contact.
302
In (11.78)a. the main clause event continues until a subsequent event occurs in a sequence of events. In
(11.78)b. the main clause event continues until a point in a story at which an event in the story takes place.
In negative main clauses the subordinate clause indicates that the non-occurrence of the event expressed in
the main clause lasts until the event in the subordinate clause takes place, effectively indicating that the
main clause event only occurs once the subordinate clause event has happened:
The elderly speaker who gave the example in (11.80) consistently used ka in this context. Among all except
the elderly, however, this usage has been completely lost, and gilai is governed directly by the sentence
head. Nonetheless, all speakers accept the presence of ka with gilai as grammatical in every instance.
Clauses governed by gilai conform to the internal structural constraints described for realis subordinate
clauses in 11.2.2.1.
Affective adverbial clauses indicate that the main clause event occurs with the intention of affecting another
participant. These clauses consist of the affective verb tufa with the affected participant as object. Affective
clauses occur with either an auxiliary or the subordinator ta.
As discussed in 7.5.2, tufa may have a benefactive or malefactive reading, depending on the semantics of
the main clause of the event.
Conditional clauses express an event or state which the main clause event is dependent on to occur. These
clauses are subordinate to the main clause, and are marked with the conditional marker la. The conditional
marker occurs immediately after the verb complex of the subordinate clause, following any postverbal
agreement enclitic or incorporated undergoer:
303
(11.82) a. [ta mai au la gai ade,]
SB come exist CND weEXC here
If we come and live here,
Unlike other subordinate clauses with the subordinator ta, the verb complex of conditional clauses may not
be marked with a cliticised demonstrative; the conditional marker and demonstrative enclitics being
mutually exclusive.
Conditional clauses typically occur sentence initially, iconically realising the temporal sequence of a
prerequisite event followed by a dependent event. However, as with legu temporal clauses (see 11.2.5.2)
this is a tendency only and the reverse order is possible, with no apparent change in meaning:
Conditional clauses are frequently introduced by the subordinator ta. As discussed in 11.2.1, ta introduces
irrealis subordinate clauses. Its occurrence in conditional clauses indicates that the event expressed by the
clause has not happened, but may yet happen. However, to use ta the speaker must have a specific event in
mind. Events which are hypothetical and not specific envisaged events are not marked with ta, nor are
positive past counterfactual conditional events. Negative past counterfactual events, however, are marked
with ta. Conditional clauses which are marked with ta are modally neutral, and may occur with the neutral
modal auxiliary g#-, although this is typically omitted:
When ta is not present in a conditional clause an auxiliary is obligatory. Ta does not occur when the
speaker regards the conditional event as very speculative and hypothetical. There is no absolute
demarcation point between future events which are regarded as sufficiently specific to be marked with ta,
and those which are speculative and hypothetical enough for the ta to be omitted. In (11.85)a., for example,
two speakers are debating the risks of custom stories falling into the wrong hands, and ta is omitted. In
304
(11.85)b. the conditional clause has the pragmatic function of a suggestion canvassing a possibility not
previously discussed:
Positive past counterfactual clauses are also not marked with ta. While all past counterfactual main clauses
and relative and complement clauses are treated as irrealis, positive past counterfactual conditional clauses
are treated as modally neutral:
Real events can also be expressed by a conditional clause, if the event is being presented as one which the
event coded by the main clause was dependent on. In this situation a realis auxiliary occurs:
As noted in 9.7.2, in one negation strategy in the language the negative existential verb teo has a sentential
complement realising the event which is negated. This is the strategy which applies to negative conditional
clauses other than past counterfactuals. As teo is the verb of the conditional clause and the subsequent
predicate its complement, the conditional marker immediately follows teo, not the negated predication:
However, conditional negative past counterfactuals are formed using the negative particle ti, in a clause
introduced by the subordinator ta:
The conditional marker may mark the existential verb au subordinated by ta to form a single word clause
translatable as 'if that is so' or 'since that is so'. This clause refers anaphorically to an event expressed by
305
the preceding main clause, introducing a main clause coding an event which is dependent on that preceding
event:
The purposive particle mala marks the event expressed by a clause as being intended or purposeful. In main
clauses mala occurs immediately following the modal auxiliary and any attached tense or aspect markers,
and preceding the verb (see 8.5.6). The particle occurs commonly, however, in subordinate clauses.
Subordinate clauses with mala indicate the purpose of the modified constituent. Purposive subordinate
clauses can function adverbially, adnominally, or as a complement clause. As adverbial subordinates they
modify an entire main clause, coding the purpose of the event expressed by main clause:
Adverbial purposive subordinates typically occur main clause finally, as in (11.91), but they may occur
initially:
Where they function adnominally, purposive subordinates indicate that the modified nominal has the
purpose expressed by the subordinate clause:
Adnominal purposive subordinates are embedded within the NP, occurring immediately after the head
nominal. Typically these clauses modify a main clause argument, as in (11.93). However, they may also
modify the nominal predicate of an equative construction:
Purposive clauses also occur as complements of a number of verbs including various verbs coding speech
events, verbs of cessation, and the desiderative manahagi. They may occur as the direct object of these
verbs, as in (11.95), or as indirect objects, as in (11.96).
306
(11.95) a. ia nahani n-e fa noto-i
theSG rain RL-3.SBJ CS stop-3SGO
The rain stopped
(11.96) manei n-e tahe-nau ara [mala tazi-ni no-na sote ana]
he RL-3.SBJ say-1SGO I PURP keep-3SGO GP-3SGP shirt thatN
He told me that I could keep that shirt of his.
As discussed in 11.2.1 above, relative and complement clauses normally occur with either an auxiliary
(when realis) or the subordinator ta (when irrealis). Purposive subordinate clauses typically occur without
either. However, in a significant minority of instances the subordinator does occur, introducing the
purposive marked subordinate. In the first line of (11.97)a. ta introduces a purposive complement clause, in
(11.97)b. an adverbial, and in (11.97)c. a relative clause.
Events coded by purposive subordinate clauses are typically either habitual (as in (11.93), (11.94),
(11.95)b., (11.96) and (11.97)b.-c.), or located in the future within the temporal frame of the main clause
(as in (11.91), (11.92), (11.95)a., and (11.97)a.). In other instances the intended event is located in the past
but did not occur. All three of these event types, habitual, future and past counterfactual, are coded as
irrealis in Kokota. Purposive subordinate clauses coding all such events are also irrealis and so could be
expected to be marked with the subordinator ta. However, as intended events are typically located in the
future (within the temporal frame either of speaking or of the main clause event), they are typically irrealis.
The presence of both the subordinator, limited as it is to irrealis clauses, and the purposive marker, is
redundant. Consequently the ta is typically omitted. However, in all such clauses its presence is optionally
possible.
The absence of auxiliaries in irrealis purposive clauses has a similar motivation. As noted in 8.5.2.5, irrealis
auxiliaries are frequently omitted. This is particularly common when mala is present, as the
307
prototypically habitual or future nature of intended events renders an irrealis auxiliary redundant. However,
they do occasionally occur. In subordinate clauses the auxiliary and mala may occur in either order. In the
second line of (11.97)a. the auxiliary precedes the purposive marker. In (11.98) it follows it:
Very occasionally a realis purposive subordinate clause occurs. As these clauses do not have the
prototypical modal status, the auxiliary is obligatory. Again the auxiliary may precede or follow mala:
In (11.99)a. the fitting of the stories on the tape is presented as being realis (and present tense), as the
addressee is telling stories which are already being recorded and some of which are already on the tape. It is
interesting that the realis subordinate clause is modifying an irrealis imperative main clause. This is
possible because the speaker is directing the addressee to carry out a future event to conform to a present
state. The example in (11.99)b. is coded realis because the umbrella is already covering the items
mentioned at the point in the temporal frame of the story.
The absence of an overt auxiliary in a purposive subordinate does not prevent the presence of a tense or
aspect marking which would otherwise be suffixed to the auxiliary, such as the perfective aspect marker in
(11.100)a., and the present tense marker in (11.100)b.:
308
11.2.7.3 Internal structure of purposive subordinates
Purposive subordinates need not be clauses with verbal predicates. Any kind of predicate can occur as the
purpose of the modified clause or nominal. For example, in (11.101) possessive predicates have been
subordinated (in (11.101)a. adverbially, in (11.101)b. adnominally):
Purposive clauses with verbal predicates only allow the language's unmarked clause constituent structure of
VAO or VS, followed by any oblique arguments. As with other clause types, it is rare for all arguments to
be specified. In (11.100)b., for example, no overt arguments occur. However, an argument with any
grammatical relation may occur. An unergative subject is present in (11.97)b.; an unaccusative subject in
(11.99)a.; direct objects in (11.99)b.; and an oblique in (11.100)a.. Multiple arguments are possible, as in
(11.98), where an unaccusative subject and two obliques receive overt mentions. As discussed in 9.3, there
is an overall tendency in the language for recently mentioned participants to receive zero mentions.
Consequently, as with other clause types, purposive clauses with all arguments are rare. However, as with
other clause types, they do occasionally occur, as in (11.95)b., where an actor and object are both present.
While any argument may occur in adverbial purposive clauses, in relative clauses of any type the controlled
argument may not be overtly realised (see 11.2.3), and this applies equally to purposive relative clauses.
Although purposive subordinate clauses do not allow any of the pragmatically marked clause constituent
orders this does not preclude clauses in which an incorporated undergoer precedes the actor. As in other
clause types, purposive subordinate clauses allow undergoer incorporation:
Purposive subordination may have scope over more than a single clause. Clause chaining may occur, with
the purposive marker having scope over the entire chained structure:
In purposive relative clauses the participant which is coreferential with the nominal head typically has a
peripheral function, such as an instrument (as in (11.93) and (11.99)b.) or locative (as the temporal
locatives in (11.97)c. and (11.100)a.). In fact mala often simply indicates that the head has a purpose which
is associated in some way with the event in the subordinate clause. This association can be simply one of
accompaniment. In (11.104) the song accompanies the activity coded by the purposive clause:
Adverbial purposive subordinate clauses may contain an argument which is coreferential with a main clause
argument. Where that is so the participant may have any grammatical relation in the subordinate clause.
309
The relation may be the same in both clauses, however it need not be - in (11.91), for example, the main
clause object participant occurs as the subordinate clause subject.
11.3 Recapping
Kokota makes frequent use of recapping strategies to link sequences of events. This occurs in all discourse
types, including conversation and narration, but is employed most frequently in exposition, where for
substantial slabs of text every sentence may commence with some recapping device. Recapping strategies
in Kokota include the use of demonstratives, prepositional oblique demonstratives, reduced clauses, and
clauses indicating completion of a recapped event.
Clause initial demonstratives occur with anaphoric reference to the event expressed by the preceding
sentence. These demonstratives occur with the sequencer ge, placing the event coded by the sentence in
which they occur within a sequence following the event coded by the preceding clause:
Oblique marked demonstratives occur consisting of the locative preposition ka cliticised to a clausal
demonstrative (discussed in 4.1.3.3). Like demonstrative recapping, these oblique demonstratives locate the
event coded by the sentence in a sequence of events, following the event coded by the preceding sentence.
However, with oblique demonstratives the relationship between the events is closer in terms of both time,
and cause and effect. Demonstrative recapping simply locates one event after another in time. With oblique
demonstratives the event coded by the sentence is presented as occurring in response to the event expressed
by the preceding sentence (not unlike the English sentence initial at that...).
Recapping oblique demonstratives typically do not cooccur with the sequencer ge, however they may do so:
la au kuru mai-di-re-n-e-ke-u
go exist be.first come-3PLO-thoseN-RL-3.SBJ-PRF-be.thus
and stopped them from coming.
Clauses may be partially repeated as a recapping strategy. These clauses are reduced by the omission of the
modal auxiliary and any other modifier, with only the verb or verbs, and optionally one or more argument,
repeated:
311
fa hage ka kame-na-re sini age
CS ascend LOC arm-3SGP-thoseN FOC SEQ
Put [them] up on his arms and then
More commonly, the predication is marked with a 'be thus' tag clause (discussed in 11.4):
The partial repetition described in 11.3.3 occurs infrequently. A more frequently used recapping strategy
employs 'completion' clauses. Completion clauses are unusual in structure. The main clause of a completion
clause consists of the verb nhigo 'be finished'. The event which is completed is expressed by a subordinated
verb complex which cannot be marked with a modal/subject auxiliary. This verb complex is the subject of
the main predication n-e nhigo 'it is finished', which it must precede. Thus in (11.114) la roh-i in line 3 and
fa blahi in line 4 are the subjects of n-e nhigo in each completion clause. However, what is unusual is that
the arguments of the subordinated verb complex coding the completed event typically occur after the main
clause predication. Thus in (11.114) line 4, ara 'I' is the subject of the subordinated predication fa blahi. A
more literal translation of this completion clause would be something like "I bless is finished...". Arguments
occasionally occur immediately following the subordinated verb complex, as shown with the object ia suga #
'the house' in (11.111) (repeating in part (11.105):
However, typically arguments occur after the main clause predication, as the subject in (11.114) line 4. In
(11.112)a. ia tan#ano 'the food' is the object of frin#he-ni 'make it'. In (11.112)b. ka-t-au-ana 'of that' is an
oblique adjunct of the predication fa ku-kumai 'cause to drink medicine'.
312
[frin#he-ni n-e nhigo ia tan#ano si-ge],
work-3SGO RL-3.SBJ be.finished theSG food FOC-SEQ
Making the food is finished and then
Completion clauses occur during the narrative or expositional description of a series of events, recapping
the event expressed by the preceding sentence and indicating that that event is completed. These occur in
narratives to indicate that one event is carried out to completion before the next event occurs:
Completion clauses occur often in exposition, particularly when a process involving a series of stages is
being described. Each stage is typically described, then a clause indicating its completion introduces the
subsequent clause:
313
11.4 'Be thus' clauses
The verb -gu ~ -u has a broad range of functions with a semantically weak predication best translated as 'be
thus'. This verb is cliticised to an auxiliary, often forming a single word clause, except occasionally with its
quotative function, when it may occur in isolation. When it does occur in isolation it is the underlying form
gu which occurs. When cliticised the initial /•/ is deleted except when the final vowel of the host is /u/.
One major function of the 'be thus' clause is as an exclamatory tag marking a constituent, usually a
complete clause, with a sense best translated as 'that's how it is' or 'that's how it was'. The form e-u, having
a zero modal component, is formally irrealis. However, e-u is semantically bleached to the point where it
has little more than an emphatic sense. Although it may mark irrealis clauses, as with the future event in
(11.115)a., e-u also marks modally neutral clauses, as in (11.115)b., and realis events ((11.115)c.):
In other instances e-u actually marks an event as irrealis. In this case the person indexing agrees with the
subject of the clause. This applies to events in the full range covered by irrealis in main clauses, including
future and habitual events, as in (11.116)a.-b., but does not apply to the irrealis example in (11.115)a., as
the subject of the main clause in that example is first inclusive. When used in this less semantically
bleached sense, the perfective aspect marker ke may mark the auxiliary, as in (11.116)c.
This less bleached use of irrealis tags is limited to third person subjects, however. The analogous first and
second person tags *a-u and *o-u do not occur.
Realis marked 'be thus' clauses always have a less semantically bleached sense than that of the irrealis tags
illustrated in (11.115). While these irrealis tags have a general sense of 'the way things are', realis tags
emphasise that the situation expressed by the clause is the way that event or state actually is or was. Realis
tags may mark realis or modally neutral clauses (as in (11.117)a.-b.), or clauses with no modal auxiliary
314
(as in (11.117)c. and the second clause in (11.117)d.). When marking a modally neutral clause or one
without an auxiliary they assign realis status to the events coded by the clause:
The less semantically bleached nature of the realis tags is indicated by their subject agreement possibilities.
The irrealis tag only occurs with the third person subject agreement marker e, regardless of the identity of
the participants in the marked constituent. By contrast, the realis tags may be indexed to a participant in the
marked clause:
In both these examples the third person tag n-e-u can freely occur in place of the tags shown, give the
slightly different sense 'It is/was like that' in place of 'I was/we are like that'.
Realis tag clauses may carry the perfective aspect marker ke and the progressive marker -gu:
7
The form fakae is given elsewhere as a monomorphemic verb. The varied status of fakae as a
monomorphemic verb and a causativised version of kae is discussed in section 3.5 of Appendix 3.
315
manei n-e-ke n#hau ge-na namhari n-e-ke-u-gu
he RL-3.SBJ-PRF eat CP-3SGP fish RL-3.SBJ-PRF-be.thus-PRG
he was eating fish, he was like that.
As (11.119)b. illustrates, tag clauses may mark verbless predicates as well as verbal clauses.
Tag clauses optionally cliticise to the preceding word. The implications of this cliticisation for stress
placement are discussed in 3.1.5.
A 'be thus' clause may occur sentence initially, to indicate that the sentence is a comment on the content of
the preceding discourse. In (11.120)a. the speaker has described two illnesses (attributed to 'devils'), and the
appropriate custom medicines for each. He concludes the exposition with the sentence given. In (11.120)b.
the speaker has explained at some length that an expected visitor is unwell, and has sent him in his place,
and concludes with the example.
Sentence initial 'be thus' clauses of this kind may be formally realis or irrealis, but only third person
indexing is possible.
Irrealis 'be thus' clauses occur as exclamatory clauses, marked with the constituent modifiers discussed in
9.8:
b. e-u baiu
3.SBJ-be.thus PSB
Maybe. [Response to proposition.]
'Be thus' clauses occur with a quotative function, assigning the origin of remarks or thoughts to the sole
argument of the clause.8 These 'be thus' clauses immediately follow a representation of the comments or
thoughts. Often comments are presented as reported speech and directly quoted:
Quotative 'be thus' clauses are often marked with the immediate marker n#a. They are typically modally
neutral, however realis and irrealis quotative clauses do occur:
8
An identical form in Zabana is analysed by Fitzsimons (1989) as a verb 'to say'. His examples suggest it
may in fact be limited to a quotative function like that in Kokota.
316
(11.123) a. 'ago' n-e-u-n#a manei
youSG RL-3.SBJ-be.thus-IMM he
"You!" He said.
With this quotative function only third person agreement occurs, regardless of the person identity of the
subject, as (11.123)b. illustrates. With this quotative function it is possible, at least among older speakers,
to omit the auxiliary entirely, giving the verb in isolation:
Quotative uses of the 'be thus' verb may be transitive, with postverbal agreement indexing the participant
to whom the comments are addressed:
b. 'nariha' n-e-ke-u-n#a-ia
day.after.tomorrow RL-3.SBJ-PRF-be.thus-IMM-PRO
"The day after tomorrow." They thought.
'Be thus' clauses are also used to assign authorship to comments or ideas without directly quoting remarks.
In this situation the 'be thus' clause is obligatorily marked with the perfective aspect marker ke, and with the
demonstrative enclitic -o, indexing the comments or ideas cited:
317